An OpenCL-accelerated parallel immunodominance clone selectionrnalgorithm for feature selection

摘要

Immunodominance clone selection algorithm (ICSA) is a robust and effective metaheuristic method forrnfeature selection problem. However, ICSA is usually slow in finding the optimal solution. In this paper,rnwe propose a parallel immunodominance clone selection algorithm (PICSA) on Graphics Processing Unitrn(GPU) to improve the speedup of ICSA for feature selection problem. The parallel program can considerablyrnaccelerate the feature selection operator. The immunodominance operator, which efficiently connects thernlocal and global information, makes the algorithm able to jump out of the local optimum easily and obtainrnthe global optimum. When comparing with other parallel languages, Open Computing Language (OpenCL)rnhas advantages both in efficiency and portability. Therefore, we use OpenCL to implement this algorithm onrnIntel many integrated core and different GPU platforms. Experiment results obtained using highdimensionalrnUCI machine learning and image texture datasets demonstrate that the PICSA algorithm allowsrnone to achieve good acceleration ratio while maintaining similar classification accuracy to serial ICSArnprogram. Besides, the OpenCL-based implementation of PICSA shows good portability on many integratedrncore and different GPU platforms as well.